Researchers from the University of Michigan, including Yonggeon Lee, Jibin Hwang, Alfred Malengo Kondoro, Juhyun Song, and Youngtae Noh, have developed a new approach to extend the life of electric vehicle (EV) batteries. Their work focuses on delaying full charging until just before departure, which can help mitigate the rapid degradation of lithium-ion batteries (LIBs) when they are kept at high states of charge (SOC) for prolonged periods.
The researchers propose a Transformer-based real-time-to-event (TTE) model for accurate EV departure prediction. This model represents each day as a TTE sequence by discretizing time into grid-based tokens. Unlike previous methods that rely heavily on historical patterns, the new approach leverages streaming contextual information to predict departures more accurately. This method is particularly useful for capturing irregular departure patterns within individual routines.
The study involved 93 users and passive smartphone data, demonstrating that the TTE model outperforms baseline models in predicting departure times. By accurately predicting when an EV will be used, the model enables delayed full charging, which can significantly extend the lifespan of the battery. This is crucial for the energy industry as it can lead to more sustainable and efficient use of EV batteries, reducing the need for frequent replacements and lowering overall costs.
The research was published in the journal Nature Communications, highlighting its potential for practical deployment in sustainable transportation systems. This innovation could be a game-changer for the energy sector, particularly in managing EV charging infrastructure and optimizing battery longevity. As the adoption of EVs continues to grow, such advancements will be essential for ensuring the long-term viability and efficiency of electric mobility.
This article is based on research available at arXiv.